Identifying the Accuracy of Clinical Diagnosis of Oral Lesions and Independent Risk Factors for Misdiagnosis

     The accuracy of clinical diagnosis of oral lesions can have a profound impact on the early detection of oral cancer. Cancer of the mouth and pharynx ranks as the sixth most commonly detected cancer in the world and is responsible for 8,000 deaths annually in the United States. Currently, the standard of care is histologic confirmation of the clinical impression for all oral lesions. While encouraging early biopsy to detect malignancies is important, it may also result in an abundance of biopsy procedures and patient worry. Specimens from common benign appearing oral lesions such as fibromas, mucoceles, and papillomas are routinely submitted for histologic evaluation, even though there may be little discrepancy between clinical and histologic diagnosis. An improved understanding of the accuracy of clinical impressions of oral lesions is necessary to inform evidence-based recommendations for biopsy and clinical care in this field.

     In this retrospective cohort study we examined the rate of discordance between clinical and histologic diagnosis of oral lesions among patients undergoing biopsy. Additionally, we investigated whether or not there are lesion- or patient-specific variables that result in a higher rate of discordance.

     This study population was drawn from patients who underwent biopsy of oral lesions from 2005-2013 by oral surgeons at The Massachusetts General Hospital. Inclusion criteria included full records, biopsy, pre-operative clinical diagnosis, and post-operative histologic diagnosis. Incomplete records resulted in exclusion. Concordance was determined by comparing the clinical impression to final histologic diagnosis. Concordance was evaluated for individual lesion types and intraoral location. In addition, lesions were stratified clinically and histologically into worrisome (malignant/pre-malignant) and non-worrisome (benign) designations, and the concordance rate of these categories was also assessed.  Factors associated with discordance were evaluated through univariate and multivariate regression analysis.

     The study sample was composed of 1003 oral lesions (pathologically confirmed 929 non-worrisome; 74 worrisome) from patients with a mean age of 44.8 years. Of the lesions evaluated, concordance between exact clinical and histologic diagnoses was found in 61% of cases. Overall, clinical diagnosis as either worrisome versus non-worrisome was 48.8% sensitive and 98.1% specific. Of note is that clinicians were able to accurately identify lesions as non-worrisome in 96% of cases. The most common of these were fibromas (Positive Predictive Value (PPV): 99.2%), mucoceles (PPV: 98.1%), and squamous cell papillomas (PPV: 96.3%).  However, the concordance for suspected worrisome lesions was only 66.7%. Several independent risk factors were associated with discordant diagnoses: radiation therapy history (p=0.0102), male gender (p=0.0381), and patient age (p=0.0468).

     To our knowledge, this is the largest study (n=1003) evaluating both clinical accuracy of diagnosis of oral lesions and patient specific factors that affect diagnosis at a tertiary care center in the United States. The results demonstrate a concerning discrepancy between clinical impression and histologic diagnosis in the worrisome lesion category (detection of 48.8%).  Non-worrisome lesions were associated with a high degree of concordance and in particular, fibromas, mucoceles and squamous papillomas were accurately diagnosed clinically. A history of prior radiation therapy, male gender and older age were all independently associated with a higher rate of discordance. These findings may help to elucidate the need for biopsy to supplement clinical suspicion, especially when the above stated risk factors are present.  Conversely, the clinician may be more comfortable with careful clinical surveillance in non-worrisome lesions in lieu of pro forma biopsy. Together, these results can help clinicians make more evidence-based decisions about the use of biopsies to confirm clinical diagnoses of oral lesions.

"Oral and Pharyngeal Cancer Statistics." National Institute of Cancer: Surveillance Epidemiology and End Results (SEER). Web. 03 Jan. 2013.

Epstein J, et al. JADA 2012; 143(12):1332-1342.